Cybersecurity Policy
Last updated: October 12, 2026
Part of the JorgAI security and trust documents.
| Document control | |
|---|---|
| Document | Cybersecurity Policy |
| Practices in effect since | Platform launch, April 2026 |
| Formal issue | October 2026 |
| Last reviewed | October 12, 2026 |
| Owner and approver | Founder, JorgAI LLC |
| Classification | Internal; shared with partners under review |
| Review cycle | Annually, and after any significant incident or architectural change |
| Related documents | Information Security Policy; Incident Response Procedure |
1. Purpose
This policy defines the technical controls that protect JorgAI's systems and the customer data they hold against unauthorized access, misuse, disruption and loss. It is the companion to the Information Security Policy, which sets governance, data handling and personnel rules; this document covers how the platform is built, run, monitored and recovered.
2. Scope
All production systems operated by JorgAI LLC: the web application and API, background workers, the realtime market scanner, databases and caches, the marketing website, the mobile applications, the continuous integration pipeline, cloud accounts and source repositories, and the vendor services that support them.
3. System overview
| Component | Technology | Location and exposure |
|---|---|---|
| Web application and API | Containerized services on Amazon ECS | Private subnets; reachable only through the load balancer |
| Background workers | Celery workers and scheduler on ECS | Private subnets; no inbound exposure |
| Database | Amazon RDS (PostgreSQL), encrypted at rest | Private subnets; not publicly accessible |
| Cache and queues | Amazon ElastiCache (Redis) | Private subnets |
| Secrets | AWS Secrets Manager | Injected into containers at start |
| Marketing website | Next.js on Vercel | Public; no customer account data |
| Mobile applications | iOS and Android clients of the same API | Public app stores; no local secrets |
| Logs and monitoring | CloudWatch, CloudTrail, Sentry | AWS and Sentry, United States |
All production infrastructure is in the United States (AWS us-east-1).
4. Network and infrastructure security
Application services run in private subnets; the only public entry point is the load balancer, which terminates TLS. The database and cache have no public endpoint and accept connections only from the application security group. There is no SSH exposure; operational tasks run as short-lived containers inside the private network using the same images as production. Development and test environments are separate from production and contain no customer data. Infrastructure changes are made through the AWS console or CLI under named identities and are recorded in CloudTrail.
5. Identity and access management
| Control | Implementation |
|---|---|
| Cloud identities | Named AWS IAM identities with least-privilege policies; multi-factor authentication required; the routine operations identity cannot modify IAM |
| Deployment credentials | Short-lived credentials issued to the CI pipeline through OpenID Connect; no long-lived deploy keys |
| Source control | Private repositories; multi-factor authentication required; protected main branch |
| Application administration | Separate admin role; step-up authentication for sensitive actions; all admin actions audited |
| Customer authentication | Email verification; bcrypt password hashing; rate-limited sign-in; optional multi-factor authentication; server-side session revocation |
| Review | Quarterly access review and on any personnel or vendor change |
6. Secrets and encryption
All application secrets, vendor credentials and the data encryption key live in AWS Secrets Manager and are injected into containers at start. None are stored in source code, container images or on developer machines, and repositories are checked for secret patterns. Customer brokerage tokens and other restricted fields are encrypted with authenticated symmetric encryption before they are written to the database, with the key held separately in Secrets Manager. Storage, backups and snapshots are encrypted at rest. All traffic, internal and external, uses TLS. Secrets are rotated on suspected exposure, on personnel or vendor change, and on a schedule; rotation is logged and new values are deployed through the pipeline.
7. Application security
The application enforces per-customer data scoping on every query, so one customer's records are never returned to another. Input is validated at the API boundary. Browser sessions are protected with a content security policy, standard security headers and short-lived access tokens with revocable refresh tokens. Broker connections use each broker's OAuth flow with revocable tokens, or API keys the customer creates; JorgAI never stores brokerage passwords and never uses unofficial APIs. Automated trading acts only within limits the customer sets, including position size, daily trade caps, stop-losses and an optional approval step for every buy, and the software has no ability to withdraw funds. Automated tests covering authentication, authorization, broker handling and the trading rules run on every change before deployment.
8. Vulnerability and patch management
Dependencies and base images are monitored for published vulnerabilities and updated promptly, with critical and high severity items prioritized and deployed through the normal pipeline. An internal security review covering the application, website, cloud configuration and dependencies was completed in October 2026; critical findings were remediated within one day and the remaining items are tracked to closure with owners and dates. JorgAI intends to commission an independent penetration test and will remediate high and critical findings before closure and share the results with partners under review.
9. Logging, monitoring and alerting
| Source | What is recorded | Retention and use |
|---|---|---|
| Application audit log | Sign-ins, broker connections and disconnections, settings changes, administrative actions, attributed to the acting identity | Retained in the database; reviewed in incidents and quarterly |
| Container and infrastructure logs | Application and worker logs, access logs | CloudWatch; reviewed in incidents |
| Error monitoring | Application exceptions with stack context | Sentry; alerts to the founder |
| Cloud API activity | All AWS API calls | CloudTrail |
| Trading activity | Every automated decision, including skipped trades and the reason | Application database; visible to the customer in their activity feed |
Logs exclude tokens, passwords and encryption keys. Security-relevant events, failed sign-in patterns and error spikes alert the founder.
10. Change management and secure deployment
- Changes are made on a development branch in version control.
- The continuous integration pipeline runs the automated test suite and the frontend build; a failing suite blocks the deploy.
- Merges to the protected main branch trigger the pipeline, which builds versioned container images and deploys to production using short-lived credentials.
- Database migrations are applied before the code that depends on them.
- Rollback is a redeploy of the previous release.
- No manual change is made to production outside this process except during an incident, in which case it is logged in the incident record and reconciled afterward.
11. Backup and recovery
| Objective | Target | Mechanism |
|---|---|---|
| Recovery time objective (RTO) | 4 hours | Redeploy versioned application images; restore the database from automated backups |
| Recovery point objective (RPO) | 5 minutes | RDS point-in-time recovery with five-minute granularity, plus daily automated backups |
| Backup testing | When the backup configuration changes and at least annually | Restore to a non-production instance and verify |
Backups are encrypted and held in the same region under the same access controls as production.
12. Incident response
JorgAI follows its Incident Response Procedure. In summary: contain within hours (revoke affected broker tokens, delete them from JorgAI systems, rotate the encryption key and any exposed credentials, invalidate affected sessions, pause automated trading for affected accounts); assess scope from the audit log, CloudWatch, Sentry and CloudTrail; notify affected partners in writing within 24 hours of confirming unauthorized access to their customers' information and provide updates until closure; notify affected customers without unreasonable delay and within applicable law, with a 72-hour target; restore service within the recovery objectives above; and deliver a written post-incident report with root cause and corrective actions within 10 business days.
13. Third-party and broker connections
Vendors are limited to those the service requires, are listed in the Information Security Policy, and are reviewed before use and annually. Broker connections use only each broker's sanctioned API and authorization flow, with scoped, revocable tokens. Partner notification commitments for outages and breaking changes are kept in the relevant partner agreements: outage notice within four hours of detection, five business days for maintenance that could interrupt order placement, and thirty days for breaking changes.
14. Endpoint and administrative device security
Devices used to administer JorgAI systems use full-disk encryption, automatic screen lock, current operating system updates and endpoint protection. Production data is never stored on them. Administrative sessions use multi-factor authentication and are closed when not in use.
15. Review and approval
This policy is reviewed at least annually and after any significant incident or architectural change. Approved by the founder of JorgAI LLC, October 2026.
| Date | Change |
|---|---|
| April 2026 | Security practices adopted at platform launch (encryption of broker credentials, OAuth-only broker access, private network, pipeline deploys) |
| October 2026 | Practices consolidated into this written policy; incident response timelines and partner notification commitments added after the October security review |
Appendix A: Control mapping
| Function | JorgAI control | Section |
|---|---|---|
| Identify | System inventory; data classification; vendor list | 3; Information Security Policy 5 and Appendix A |
| Protect | Least-privilege access with MFA; encryption in transit and at rest; secrets management; secure development; endpoint rules | 5, 6, 7, 10, 14 |
| Detect | Audit log; CloudWatch; Sentry; CloudTrail; alerting | 9 |
| Respond | Incident Response Procedure with 24-hour partner notice | 12 |
| Recover | Backups, point-in-time recovery, versioned releases, RTO 4 hours, RPO 5 minutes | 11 |